Minutes for 2013-10-29
Attending: Nick, Steve, Mike
Nick asked about LDP containers; Steve clarified that a link header is defined to point to the non-member-resources resource, and that clients can use an OPTIONS request to get the link headers - cheaper than a HEADER request, since the server does not need to calculate ETAGs, etc. As a result, OSLC Config Mgt implementations can determine which is the better approach: a config IS an LDPC, or a config POINTS TO an LDPC.
We then discussed the two different RDF representations of versioned data: triples with the concept resource as the subject, embedded in a graph which identifies the verion resource, or triples that have the version resource as the subject. Nick explained that OSLC SCM 1.0 had taken the latter approach, and that had proven problematic in IBM’s implementation - it made it more difficult to adopt versioning, since every link had to change, and it made it more difficult to write queries that were automatically scoped to a configuration. It also did not match the view that a version is just a state of the resource, and did not change the representation of the properties of that resource. Mike mentioned that some PLM tools use the version-centric representation, so an open world has to handle both. We will probably need a way for clients to discover which version representation is in use (property on a shape?) and we agreed this would be one way to help address the issue of compatibility with / migration for OSLC SCM 1.0.
We then talked briefly about the move to OASIS and the timeframe for participation agreements. Steve mentioned that we hope to have the OASIS TC Charter proposal by the end of the year, so the original proposers should be OASIS members by then. Other group members can be added later, perhaps in 1st quarter 2014.
